Questions: Event Semantics and Thematic Structure

Question 1

In standard Montague semantics without event variables, the sentence 'John sang beautifully in the park' creates a problem. Which of the following best describes that problem?

• A. The verb 'sang' cannot take an adverb as a type-theoretic argument in intensional logic
• B. To capture all the entailments (that John sang, that he sang beautifully, that he sang in the park), one must stipulate an unlimited family of distinct predicates with no principled connection between them
• C. Adverbs like 'beautifully' lack truth conditions because they express subjective evaluation
• D. Prepositional phrases are adjuncts and adjuncts carry no semantic content in Montague grammar

Answer: (B)

Without event variables, 'sang beautifully in the park' becomes a single complex predicate SANG-BEAUTIFULLY-IN-THE-PARK(John), which has no principled relationship to SANG(John) or SANG-BEAUTIFULLY(John). To capture all the entailments, you would need an infinite inventory of related predicates with stipulated implication relations between them — and no principled account of why they entail each other. Davidson's event variable solution collapses this: SANG(e) ∧ Agent(e, John) ∧ BEAUTIFULLY(e) ∧ IN-THE-PARK(e), and each entailment follows by conjunction elimination. The adverb problem becomes a straightforward consequence of predicate logic.

Question 2

In the neo-Davidsonian framework, the sentences 'The chef melted the butter' and 'The butter melted' share an underlying event structure, differing only in:

• A. The Patient/Theme argument — different objects undergo melting in each sentence
• B. The presence or absence of a Causer argument — the causative form adds Causer(e, the chef) to the same underlying melting event
• C. The tense and aspectual structure — the transitive implies completion, the intransitive implies process
• D. The thematic role assigned to 'butter' — it is a Theme in the intransitive but a Patient in the causative

Answer: (B)

This is the causative-inchoative alternation. Both sentences describe a melting event MELT(e) with Theme(e, the butter). The causative form adds Causer(e, the chef) to an event structure that is otherwise shared. In the neo-Davidsonian representation — where thematic roles are separate predicates rather than fixed argument positions — this addition or removal of a role is a systematic operation, not a stipulated lexical alternation. The insight is that what appears to be two different verbs is actually one event type with a variable argument structure, predicting the alternation rather than listing it.

Question 3

In event semantics, adverbs like 'slowly' and 'in the park' are predicates over event variables rather than modifiers of the main verb predicate, which allows adverbial entailments to follow as logical consequences rather than stipulations.

• T. True
• F. False

Answer: True

This is Davidson's core insight. 'John ran slowly in the park' becomes RAN(e) ∧ Agent(e, John) ∧ SLOWLY(e) ∧ IN-THE-PARK(e). The entailment 'John ran slowly' follows by conjunction elimination — drop the IN-THE-PARK(e) conjunct. Without event variables, you have a single complex predicate RAN-SLOWLY-IN-THE-PARK(John) with no sub-parts to eliminate and no principled account of why it implies RAN-SLOWLY(John). The event variable is what transforms an intuition about entailment into a formal consequence.

Question 4

Event variables in event semantics are simply a notational shorthand — they do not add semantic content beyond what was expressible in standard Montague semantics without events.

• T. True
• F. False

Answer: False

Event variables are genuine semantic objects with their own expressive power that goes beyond notation. They can be quantified over ('John kissed Mary twice' quantifies over kissing events), referred to by demonstratives ('That surprised me' picks up a previously mentioned event), and located by temporal expressions ('before noon' locates an event in time). None of these constructions are tractable in standard Montague semantics without events. The framework is not a notational variant of what came before — it extends the descriptive range of semantic theory to cover quantification over events, event reference, and temporal modification in a principled way.

Question 5

Why does the neo-Davidsonian decomposition of thematic roles into separate predicate positions provide a better account of argument alternations than the original Davidsonian representation?

Model answer: In the original Davidsonian representation, GIVE(e, x, y, z) bundles Agent, Theme, and Goal into fixed argument positions. To handle alternations like passivization or ditransitive variants, separate lexical entries with stipulated relationships would be needed. In the neo-Davidsonian version — GIVE(e) ∧ Agent(e, x) ∧ Theme(e, y) ∧ Goal(e, z) — each thematic role is an independent predicate that can be suppressed, demoted, or rearranged. Passivization suppresses the Agent predicate from surface syntax while leaving the event structure intact. The causative adds a Causer predicate to an otherwise unchanged event. These become systematic operations rather than language-specific stipulations.

The separability of thematic roles is not just a formal nicety — it makes testable predictions. If roles are truly independent, we expect the same role to behave similarly across different verbs and alternations, and we expect the suppression or addition of a role to have consistent syntactic effects across the lexicon. This transforms what was a collection of observed alternations into derived predictions of a principled theory, which is the mark of genuine theoretical progress in semantics.